JP6668321B2 - Error compensation apparatus and method - Google Patents

Description

  The present invention relates to a type of error compensating device, and more particularly to an error compensating device that can be applied under different cutting conditions through a type of module for constructing coefficients in advance.

  Since the Third Industrial Revolution, PLC robot arms and robots have begun to occupy an important position in industrial production. In particular, for robot arms widely used in precision machining and the IT industry, the positioning accuracy is extremely important.

  For a robot arm, the difficulties in positioning are where the need to overcome various inevitable errors can be broadly divided into two types: geometrical errors, such as parameter errors for each rod of the robot arm. , There is an error between the reference coordinates and the actual spatial coordinate system, or an error in the parallelism of the axis of each joint of the robot arm. The second is non-geometric error. Among them, the most prominent is a thermal error, and the next is a backlash error generated when combining gears, or a deformation error due to external force of a joint or a connecting rod or its own weight.

  Since the above-mentioned errors appear on the robot arm as a whole, it is difficult to construct an error compensation module that can deal with various error sources one by one. Even if an error compensation module is constructed, there are individual errors in each part of the robot arm, or errors generated during assembly are different from each other, so that the error compensation module often cannot be applied. This makes the error compensation calibration work unusually difficult.

  Therefore, in order to overcome the deficiency of the existing technology, the present invention defines an absolute error using the amount of position movement of an undeformed speckle on the object surface, and performs a compensation operation based on the error. And how to submit.

  According to one embodiment of the present invention, a kind of error compensating device is provided. The error compensating device includes at least one camera module, one matching module, and one calibration compensation module. The camera module constructs one spatial coordinate system for the aforementioned objects, and the camera module includes a first camera device and a second camera device. The first camera device captures a plurality of first non-deformed speckle images corresponding to the first surface of the object and on the first surface. The second camera device captures a plurality of second non-deformed speckle images corresponding to the second surface of the object and on the second surface. The first and second surfaces are one-sided with respect to the object. The matching module matches the two first non-deformed speckle images and the two second non-deformed speckle images before and after the position movement of the first surface and the second surface, respectively, and based on the first surface and the second surface. The position movement amount of the two surfaces in the space coordinate system is calculated. The calibration compensation module is connected to the object, and controls the movement of the object based on the two movement amounts after the matching module receives the movement amount of the first surface and the second surface.

  Accordingly, the method of the present embodiment can determine the movement distance and direction of any one point on the object surface through the collation of non-deformed speckle images photographed before and after the position movement of the first surface and the second surface. . In addition, since the position shift amount of the speckle is already a comprehensive expression result of all errors, it is not necessary to construct a compensation module again for various error causes, and has a faster and more accurate error compensation effect. .

  In one embodiment, the spatial coordinate system may be a rectangular coordinate system or a cylindrical coordinate system.

  In one embodiment, the position shift amount can be a length or an angle.

  In one embodiment, the azimuth may be 90 degrees.

  In one embodiment, the above objects can be placed on one robot arm.

In one embodiment, the object may be Invariable Steel (Invar® ) or a zero-expansion glass-ceramic material.

  In one embodiment, the position shift amount of the first surface and the second surface is a sum of absolute error values (Sum of Absolute Differences, SAD), a square error (Sum of Squared Differences, SSD), and a robust feature amount. The calculation is performed by speeding up (Robust Features, SURF), scale-invariant feature transform (SIFT), or normalized cross-correlation (NCC).

  In one embodiment, the number of the camera modules is two, and the two corresponding camera modules are located on both sides of the corresponding object.

According to the above-described embodiment, the present invention can be applied to the task of compensating for errors in rectangular coordinates or cylindrical coordinates, and the two camera devices take pictures from different azimuth angles, so that the object can be photographed in all three-dimensional directions. The movement can be calculated accurately. In addition, the use of Invariable Steel (Invar® ) or zero-expansion glass allows the error compensator to be maintained in a calibrated state, and a non-deformable spec that the camera module captures by moving its own position. The problem that the actual image does not correspond to the actual image can be avoided.

  According to another embodiment of the present invention, there is provided a kind of error compensation method. The method includes: providing at least one camera module including a first camera device and a second camera device. Construct one spatial coordinate system for camera module objects. The first camera device is operated to capture a plurality of first non-deformed speckle images on the first surface of one of the aforementioned objects. Operate the second camera device to take a plurality of second non-deformed speckle images on the second surface of one of the aforementioned objects, wherein the first surface and the second surface have one-sided difference with respect to the aforementioned objects. There is. The two first non-deformed speckle images and the two second non-deformed speckle images are compared before and after the first surface and the second surface are moved. Calculating the position movement amount of the first surface and the second surface in the spatial coordinate system based on the two first non-deformed speckle images and the two second non-deformed speckle images. The movement of the object is controlled based on the position movement amount described above.

  According to the above-described error compensation method, the present invention captures an undeformed speckle image through a camera device, and then compares the non-deformed speckle image taken before and after the movement of the object, thereby obtaining any surface on the object. , The amount of movement at any position can be quickly obtained, and the amount of movement can be directly applied to error compensation. Therefore, the error compensation method employed in the present embodiment does not need to be calibrated for various error causes one by one, and directly calculates the position movement amount of the photographing position, thereby making the error compensation more accurate and efficient. .

  In one embodiment, the spatial coordinate system may be a rectangular coordinate system or a cylindrical coordinate system.

  In one embodiment, the position shift amount can be a length or an angle.

  In one embodiment, the azimuth may be 90 degrees.

  In one embodiment, the error compensation method described above can include: placing the object on one robot arm.

In one embodiment, the object may be Invariable Steel (Invar® ) or a zero-expansion glass-ceramic material.

  In one embodiment, the position shift amount of the first surface and the second surface is a sum of absolute error values (Sum of Absolute Differences, SAD), a square error (Sum of Squared Differences, SSD), and a robust feature amount. The calculation is performed by speeding up (Robust Features, SURF), scale-invariant feature transform (SIFT), or normalized cross-correlation (NCC).

  In one embodiment, the number of the camera modules is two, and the two corresponding camera modules are located on both sides of the corresponding object.

FIG. 2 is a structural block diagram of an error compensating device according to one embodiment of the present invention. FIG. 2 is an image diagram of a camera module of the error compensating device of FIG. 1 in a rectangular coordinate system. FIG. 2B is an image diagram of a position movement of a non-deformed speckle image of the error compensating device of FIG. FIG. 2 is an image diagram of a camera module of the error compensating device in FIG. 1 in a cylindrical coordinate system. FIG. 3B is an image diagram showing a position movement of a non-deformed speckle image of the error compensating apparatus of FIG. FIG. 2 is an image diagram when the error compensation device of FIG. 1 is applied to a robot arm. FIG. 4 is a step process diagram of an error compensation method as another embodiment of the present invention.

In order to fully understand the objects, features and effects of the present invention, the present invention will be described in detail with reference to the following specific examples and the accompanying drawings. The description is as follows:
Please refer to FIG. 1, FIG. 2A and FIG. 2B. The error compensating device 100 of the present embodiment is used for detecting and compensating for a position movement error of one object O, and the error compensating device 100 includes at least one camera module 200, one matching module 300, and one calibration. A compensation module 400 is included. The camera module 200 constructs one spatial coordinate system for the aforementioned object O, and the camera module 200 includes one first camera device 210 and one second camera device 220. As shown in FIGS. 2A and 2B, the first camera device 210 corresponds to the first surface E1 of the above-described object O, and the plurality of first non-deformed speckle images P1, P1 ′ on the first surface E1. To shoot. More specifically, when the above-described object O moves, the position of the first surface E1 corresponding to the first camera device 210 changes, and thus the first non-deformed speckle image P1 captured by the first camera device 210. Also changes its position to P1 '. Therefore, the above-mentioned plurality of first non-deformed speckle images P1 and P1 ′ mean a set of images captured by the first camera device 210 for each position on the first surface E1 when the object O moves. . Similarly, the second camera device 220 captures a plurality of second non-deformed speckle images P2 and P2 ′ corresponding to the second surface E2 of the object O and on the second surface E2. Among them, the first surface E1 and the second surface E2 have a difference in one-sided angle with respect to the object O. When the position of the first surface E1 and the second surface E2 is moved, the collation module 300 performs two first non-deformed speckle images P1 and P1 ′ before and after the position movement and two second non-deformed speckle images. , Respectively, thereby calculating the position movement amount d of the first surface E1 and the second surface E2 in the aforementioned spatial coordinate system.

  In the present embodiment, the non-deformed speckle image is formed by irradiating the surface of the object O with a coherence light source and acquiring light rays emitted from a pattern on the surface of the object O. Under the condition of the microscopic viewpoint, the patterns captured at any two points on the surface of the object O cannot completely match. Therefore, a database of each image and position is created based on the pattern images of each position on the first surface E1 and the second surface E2 taken in advance, and is applied to the subsequent error compensation work. For example, the first camera device 210 captures two positions on the first surface E1 before and after the movement of the object O, and the matching module 300 generates two first non-deformed speckle images P1 and P1 ′. By collating with the database, it is possible to quickly find the positions of the two first non-deformed speckle images P1 and P1 ′ on the first surface E1, and calculate the position movement amount d of the first surface E1.

  The position movement amount d described in this embodiment method is a sum of absolute error values (Sum of Absolute Differences, SAD), a square error (Sum of Squared Differences, SSD), and a robust feature amount (Speeded Up Robust Features, SURF). , Can be calculated by Scale Invariant Feature Transform (SIFT) or Normalized Cross Correlation (NCC), but is not limited thereto.

  Using the method described above, the position of the undeformed speckle image before and after the movement is acquired, and the calibration compensation module 400 receives the above-described position movement amount d and performs an error compensation operation. More specifically, all the error compensation operations use forward kinematics and inverse kinematics to correct errors repeatedly, and the forward kinematics calculates the amount of movement, rotation, or The final position of any one point on the surface of the object O is calculated based on the position movement due to other causes. Inverse kinematics determines the movement method of the object O based on the current position of any one point on the surface of the object O and the next target position. In the present embodiment, the calibration compensation module 400 is connected to the object O, and when an error occurs, the calibration compensation module 400 first uses the inverse kinematics to calculate the position based on the position movement amount d. A movement parameter for correcting the movement amount d is calculated (for example, a linear movement amount and a rotation amount), and then the object O is moved to the target position. When the first surface E1 or the second surface E2 reaches the new position, the matching module 300 checks the difference between the new position and the target position, and adjusts the above-described movement parameter using forward kinematics. Then, the new position is compared again with the position movement amount d of the non-deformed speckle image at the target position, and based on that, the calibration compensation module 400 continuously compensates for the error.

  Please refer to FIG. 3A and FIG. 3B. In other embodiments, the object O to which the camera module 200 can be applied is not limited to a flat surface. For example, the object O may have a cylindrical or other irregular shape, and the spatial coordinate system may be a rectangular coordinate system (ΔX, ΔY, ΔZ), a cylindrical coordinate system (Δr, Δθ, Δh) or another coordinate system. There may be. Similarly, the position movement amount d is not limited to the length, but may be an angle.

  As shown in FIGS. 2B and 3B, the difference between the azimuth angles of the first surface E1 and the second surface E2 can be 90 degrees. In FIG. 2B as an example, the first camera device 210 is limited to the shooting direction, and it is difficult to shoot the position movement amount d in the Z direction of the first non-deformed speckle image, and the second camera device 220 2. It is difficult to capture the position movement amount d of the non-deformed speckle image in the X direction. Therefore, according to the azimuth difference between the first surface E1 and the second surface E2, the present embodiment method can determine the position movement amount d of the object O from different directions. The point to be particularly described is that the azimuth of 90 degrees is for obtaining the amount of position movement d in each axis direction of the coordinate system, and when the surface of the object O is irregular (for example, when the object O is Or installed on the robot arm M), the azimuth angle is likely to be larger or smaller than 90 degrees. However, the difference in the azimuth does not affect the calculation of the position movement amount d.

Please refer to FIG. The drawing is an image diagram when the error compensating apparatus 100 of the present embodiment is applied to the robot arm M. The error compensating apparatus 100 may include two or more camera modules 200, and the positions of the two camera modules 200 correspond to both sides of the object O, respectively. As shown in FIG. 4, since the robot arm M is longer, its own weight may cause a non-geometric error at the distal end. By disposing the camera modules 200 on both sides of the object O, it is possible to more accurately measure the position movement amount d of the first surface E1 and the second surface E2. In addition to directly installing the error compensating device 100 on the robot arm M, the error compensating device 100 can be installed on an object O of a zero-loading frame (unloading frame), and unexpected deformation of the object O itself may occur. To avoid, the aforementioned object O can be made of Invariable Steel (Invar® ) or a zero-expansion glass-ceramic material.

  FIG. 4 shows an example of a robot arm M having a multi-stage connecting rod. The camera module 200 can be installed on the object O or the robot arm M, and the error compensation device 100 can Can be included. With the pedestal of the robot arm M as a reference store, the position movement amount d of all camera modules 200 is used to calibrate the position of the next camera module 200. It is possible to calibrate the position movement amount d. Thus, after the position error of each camera module 200 on the robot arm M is obtained, calibration compensation is performed for each part of the robot arm M up to the actuator at the end of the robot arm M one by one.

  According to the above implementation method, the implementation method can be applied to the error compensation of the robot arm M or other automation equipment, and the implementation method performs surveying and compensation on the final result of the error, so that various error sources can be obtained. In addition, it is not necessary to consider the correction method individually, so that not only is the compensation result more accurate, but also the difficulty of error compensation is reduced.

  Please refer to FIG. According to another embodiment of the present invention, a kind of error compensation method 500 is provided. The method includes the following steps: Step 510 provides at least one camera module 200 including one first camera device 210 and one second camera device 220. Step 520 constructs one spatial coordinate system for the object O of the camera module 200. In step 530, the first camera device 210 is operated to photograph a plurality of first non-deformed speckle images P1 and P1 'on one first surface E1 of the object O. Step 540 operates the second camera device 220 to capture a plurality of second non-deformed speckle images P2 and P2 ′ on one second surface E2 of the object O, of which the first surface E1 and the second surface E2 are There is a one-sided angle difference with respect to the object O. Step 550 compares the two first non-deformed speckle images P1 and P1 ′ and the two second non-deformed speckle images P2 and P2 ′ before and after the position movement of the first surface E1 and the second surface E2. Step 560 is based on the two first non-deformed speckle images P1, P1 'and the two second non-deformed speckle images P2, P2' in the aforementioned spatial coordinate system of the first surface E1 and the second surface E2. Calculate the position movement amount d. Step 570 controls the movement of the object O based on the above-mentioned position movement amount d.

  The detailed implementation method of the error compensation method 500 is the same as that described in the above-described error compensation apparatus 100, and thus will not be described again in detail. In this embodiment, the final position movement result of any position on the object O is directly measured, and calibration is performed based on the measured result. The error compensation method 500 has a faster and more accurate effect than the existing technology, for example, a method of constructing various thermal error models and force analysis models, and further performing calibration compensation one by one. At the same time, calibration is performed for any position movement results, and there is no need to consider backlash errors during component assembly and size errors of the components themselves, thus simplifying calibration compensation work. is there.

  In one embodiment, the spatial coordinate system of the error compensation method 500 described above can be a rectangular coordinate system or a cylindrical coordinate system.

  In one embodiment, the position movement amount d of the above-described error compensation method 500 can be a length or an angle.

  In one embodiment, the azimuth of the error compensation method 500 described above may be 90 degrees.

  In one embodiment, the object O of the error compensation method 500 described above can be installed on one robot arm M.

In one embodiment, the object O of the above-described error compensation method 500 can be Invariable Steel (Invar® ) or a zero-expansion glass-ceramic material.

  In one embodiment, the displacement amount d of the first surface E1 and the second surface E2 of the above-described error compensation method 500 is the sum of absolute error values (Sum of Absolute Differences, SAD), the sum of square errors (Sum of Squared Differences, SSD), speeded up robust features (SURF), scale-invariant feature transform (SIFT), or normalized cross-correlation (NCC).

  In one embodiment, the number of camera modules 200 of the error compensation method 500 is two, and the two camera modules 200 are located on both sides of the corresponding object O, respectively.

  The present invention has disclosed a superior embodiment among the above. However, those skilled in the art should understand that the embodiments are for the purpose of drawing the present invention and should not be interpreted as limiting the scope of the present invention. It should be noted that all changes and substitutions equivalent to the embodiment should be included in the scope of the present invention. Therefore, the scope of protection of the present invention shall be based on the one defined by the patent scope application.

100 Error Compensation Device 200 Camera Module 210 First Camera Device 220 Second Camera Device 300 Matching Module 400 Calibration Compensation Module 500 Error Compensation Method 510-570 Step d Positional Movement E1 First Surface E2 Second Surface P1, P1 ′ First Non-deformed speckle images P2, P2 'Second non-deformed speckle image M Robot arm O Object

Claims (10)

A kind of error compensator, which is used for detecting and compensating for an error in movement of a position of an object, the error compensator includes:
At least one camera module, which constructs one spatial coordinate system for the object, and the camera module includes:
One first camera device, corresponding to a first surface of the object, capturing a plurality of first non-deformed speckle images on the first surface, and one second camera device, a second surface of the object Corresponding to, taking a plurality of second non-deformed speckle images on the second surface, and the first surface and the second surface have a difference in one angle with respect to the object,
One database, save the pattern image of each position on the first surface and the second surface taken in advance,
One collation module, when the first surface and the second surface are moved, the collation module, the two first non-deformed speckle images and the second non-deformed speckle images before and after the position movement , The pattern image and the position stored in the database are collated with each other, and the amount of position movement of the first surface and the second surface in the spatial coordinate system is calculated based on the pattern image and the position. Connected, the calibration compensation module controls the movement of the object based on the position movement amount after receiving the position movement amount from the matching module.   In the error compensating device described in claim 1, the position movement amount is a length or an angle.   In the error compensating apparatus according to claim 1, the object is set on one robot arm.   In the error compensator according to claim 1, the object is made of invariable steel or zero expansion glass ceramic material.   In the error compensation device described in claim 1, the number of the camera modules is two, and the two camera modules are respectively located on both sides of the object. A kind of error compensation method, which is used for detecting and compensating for an error in movement of a position of one object, the error compensation method includes:
Providing at least one camera module including a first camera device and a second camera device,
Build one spatial coordinate system for camera module objects,
Operate the first camera device to capture a plurality of first non-deformed speckle images on one first surface of the object,
Operate the second camera device to capture a plurality of second non-deformed speckle images on one second surface of the object, wherein the first surface and the second surface are one-sided with respect to the object. There is a difference,
The first surface and the pattern image of each position on the second surface are photographed in advance and saved in a database,
The two first non-deformed speckle images and the two second non-deformed speckle images before and after the position movement of the first surface and the second surface, and the pattern image and the position stored in the database. Collation,
Calculating the amount of position movement of the first surface and the second surface in the spatial coordinate system based on the two first non-deformed speckle images and the two second non-deformed speckle images; and Controls object movement based on volume.   In the error compensation method described in claim 6, the position movement amount is a length or an angle. The error compensation method according to claim 6 includes:
The object is set on one robot arm.   In the error compensation method as set forth in claim 6, the object is invariant steel or zero expansion glass ceramic material. In the error compensation method described in claim 6, the number of the camera modules is two, and the two camera modules are respectively located on both sides of the corresponding object.